Clinical Investigation

Prediction of pregnancies in ICSI cycles with artificial neural networks

  • Ibrahim Esinler
  • Hakan Yarali

Turk J Obstet Gynecol 2006;3(3):176-180

OBJECTIVE: To determine the success rate of Artificial Neural Network (ANN) in prediction of clinical pregnancies in intracytoplasmic sperm injection (ICSI) cycles Design: Retrospective clinical trial. Setting: Hacettepe University Faculty of Medicine, Department of Obstetrics and Gynecology, IVF Clinic Patients: Five hundred ICSI cycles reached to embryo transfer (ET) Interventions: Artificial Neural Network (ANN) was used to predict the clinical pregnancies in ICSI cycles. Main outcome measures: Success rate of ANN in prediction of clinical pregnancies RESULTS: Overall, the ANN with best performance predicted correctly the outcomes of %70 of ICSI cycles. It predicted correctly the 53% of all positive clinical pregnancies and 81% of all cycles without clinical pregnancy. CONCLUSIONS: ANN may be use to predict the pregnancy outcome of ICSI cycles. More studies with larger sample size should be carried out to support our study which is first in our country and second in the world.

Keywords: artificial neural networks, clinical pregnancy, ICSI, IVF